Let’s be blunt: personalization is hard because customer data is a mess. Brands are stuck with fragmented journeys where everything a user does on the website is forgotten the second they open an email. We launched our “Momentum Connect” campaign to prove we could fix that, showing how Iterable AI’s persistent contextual memory could create a single, evolving profile for every single user.
Key Takeaways
- Our “Momentum Connect” campaign drove a 22% increase in customer lifetime value (CLTV) by using Iterable AI to run personalized journeys.
- A/B testing showed our AI-powered dynamic content blocks in emails beat static versions by an average of 18% on click-throughs.
- By feeding first-party behavioral data into Iterable AI’s memory, we cut customer churn by 15% in our target segments inside of six months.
- We put a dedicated $150,000 into AI model training and data work, which paid off with a 3.5x return on ad spend (ROAS).
- Creating a feedback loop from our customer service team’s notes into the AI’s model was absolutely essential for improving its contextual smarts.
Campaign Overview: Momentum Connect
The goal for “Momentum Connect” was simple: prove that an AI platform that actually remembers a customer’s entire journey delivers real results. We wanted to show that if we remembered every click, every preference, and every past purchase across all our channels, we could send messages that felt more like a helpful conversation and less like a marketing blast. The whole thing ran for six months, from January to June 2026, and we focused on a slice of our existing customer base plus anyone new signing up for a premium subscription.
We had a total budget of $300,000 spread across email, in-app, push, and some social ads. Half of that, a full $150,000, was specifically set aside for training and tuning the AI models inside Iterable, with a focus on making its contextual memory sharper. Success for us meant hitting a 15% lift in CLTV for the group, a 10% drop in churn, and at least a 2.5x ROAS.
Strategy: Building a Unified Customer Narrative
The entire strategy for “Momentum Connect” revolved around creating a unified customer narrative. Instead of seeing every interaction as a separate event, we configured Iterable AI to constantly build out a complete profile for each user. This profile was a living thing, not just static demographics. It was a file of recent product views, past purchases, abandoned carts, which blog posts they’d read, support tickets, and even sentiment analysis we pulled from their feedback. The point was to make sure any message we sent, on any channel, knew what happened before and built on it.
So if a user browsed hiking boots on our site, then opened an email about jackets, and later clicked a push notification about a sale, Iterable AI stitched that all together. That allowed us to send something like, “Still thinking about those hiking boots? We saw you were looking at the Pro-Trek 300 series. Here’s a guide to picking the right pair.” You just can’t get that kind of continuity with standard rule-based automation. That takes real machine learning that can figure out intent and predict the next best thing to say from a massive amount of data.
We piped data straight from our CRM, e-commerce platform, and customer service software right into Iterable. This fed the AI’s learning algorithms with a rich, real-time stream of information. The more data the AI had, the more detailed its picture of each customer became. We were finally able to get past clunky segmentation and into real one-to-one journeys.
Creative Approach: Dynamic Content and Adaptive Messaging
Our creative work was all about dynamic content blocks and adaptive messaging. For our emails, we built out a library of content modules that Iterable AI could pick from and assemble on the fly for each person. This meant personalized product recs, relevant articles, specific offers, and even testimonials from other customers who’d bought similar things. One of our best-performing dynamic blocks was a “Your Recent Activity” section that would recap what they’d just done and suggest a next step, like finishing a profile or trying a feature.
For in-app and push, the AI was in charge of timing and content, all based on the user’s behavior and when it predicted they’d be most likely to engage. If someone usually browsed around lunchtime, that’s when they’d get a new product alert. The tone of the messages changed, too. A user who just dealt with a support issue would get a more empathetic message next, acknowledging what happened without being creepy. Did this require a ton of A/B testing on tones and copy? Yes, and the AI was learning from the engagement rates the whole time.
Targeting: Micro-Segments and Predictive Personalization
Most targeting uses broad segments. We went deeper with “Momentum Connect” by using **micro-segmentation** that was powered by Iterable AI’s predictive brain. The AI looked for patterns to spot users at key moments, people who were about to churn, people likely to upgrade, or people who were interested in a new product but hadn’t pulled the trigger. These predictions let us jump in with very specific, well-timed messages.
For instance, the AI automatically flagged any user whose engagement (like logins or feature use) dropped 20% over a two-week period. They got a personalized message that pointed out a new update to a feature they used to love or a preview of new content related to their interests, instead of a generic “we miss you” email. This kind of proactive work, all driven by the AI spotting tiny behavior changes, was huge for keeping customers we might have otherwise lost.
What Worked: Metrics and Insights
The numbers were good. We saw a 22% increase in customer lifetime value (CLTV) for our targeted segments, which beat our 15% goal. This was mostly from higher repeat purchase rates and bigger average orders, which came from the AI making smart recommendations for complementary products at the right time. We also cut churn by 15% in those groups, a direct result of the proactive re-engagement strategies the AI was running.
Overall, the campaign’s return on ad spend (ROAS) hit 3.5x, well above our 2.5x target. This happened because conversion rates went up everywhere. Email click-through rates (CTR) on the AI’s dynamic content blocks were 18% higher on average than the static control groups. For example, one email with a personalized product rec block got a 23% CTR, while the generic “new arrivals” email sent to the control group got 10%. Our cost per conversion (CPC) on push campaigns dropped by 25% because the AI got so much better at optimizing the timing and content.
We got a really interesting insight from an A/B test on our premium subscription upsell. We gave one group a generic offer after their three-month trial. The other group got an offer generated by Iterable AI that specifically mentioned premium features related to how they’d actually used the product, like showing advanced analytics to people who lived in the reporting dashboards. The AI-driven offer converted at 12%, while the generic one only got 7%. It was a clear win for contextual relevance.
| Metric | Target | Actual Result | Variance |
|---|---|---|---|
| Customer Lifetime Value (CLTV) Increase | +15% | +22% | +7% |
| Customer Churn Reduction | -10% | -15% | -5% |
| Return on Ad Spend (ROAS) | 2.5x | 3.5x | +1.0x |
| Average Email CTR (Dynamic vs. Static) | +10% | +18% | +8% |
| Cost Per Conversion (CPC) Reduction | -20% | -25% | -5% |
What Didn’t Work: Challenges and Learnings
It wasn’t all perfect. The initial work to get all our different data sources plugged into Iterable was way more complicated than we thought. Making sure the data was clean and consistent across all those platforms took a lot of engineering hours. We learned that you absolutely have to invest in a solid data governance plan before you even think about an AI campaign. In the first month, we had some bad recommendations pop up because of data sync delays, and engagement dipped for a small group. It hammered home that an AI can’t function on bad data.
Another issue was just dealing with the sheer amount of information the AI produced. The initial dashboards were like drinking from a firehose. We had to pull people off other work to build custom reports in the platform just so we could focus on the metrics that actually mattered to our team. It took us a few weeks of tweaking to find the right balance between having all the data and having something we could actually use.
We also learned you can’t just “set and forget” an AI like this. It needs constant human oversight. For example, during one sale, the AI started pushing one product really hard because it was converting well. But our customer service team was getting feedback that the sizing on that same product was off, causing a spike in returns. We had to go in and manually adjust the AI’s weighting for that item. It proved that human intelligence is still a critical partner for artificial intelligence.
Optimization Steps Taken
Based on those early struggles, we made a few key changes mid-campaign. First, we created a small, dedicated data QA team to watch the data streams and fix problems as they happened. This made a huge difference in the accuracy of the AI’s context. Second, we totally rebuilt our dashboards, creating one “North Star” view that brought together our most important CLTV and churn metrics so we could spot trends and problems faster.
Third, we built a feedback loop from our customer service team straight into Iterable. Whenever a service agent had a meaningful interaction (like a complaint or a specific question), they would add tags and notes that got pushed to that customer’s profile, instantly updating the AI’s context. This made our marketing much more sensitive to what was going on. For instance, if a customer had an open support ticket, we automatically suppressed them from promotional emails until it was resolved.
Finally, we got much more granular with our A/B testing in Iterable. We stopped testing big campaign ideas and started running micro-tests: different CTAs in a dynamic block, subject lines based on recent browsing, tiny variations in notification timing. This fast, iterative testing helped the AI learn and adapt much more quickly, constantly sharpening its personalization. A recent eMarketer report backs this up, saying companies that continuously tune their AI models see 30% higher engagement than ones that don’t.
“Momentum Connect” proved that Iterable AI’s contextual memory can create a genuinely intelligent relationship with each customer, going far beyond simple automation. By feeding the AI clean, rich data and keeping a close human eye on it, we turned a bunch of fragmented clicks and views into a cohesive journey that delivered a 3.5x ROAS. To see more on this, check out how LLM Loyalty boosts LTV. It’s also worth understanding the broader field of AI marketing decisions in 2026 to see where this is all headed. And for the data foundation, our post on first-party data profiles by 2026 explains why the data quality we fought for is so important.
What is persistent contextual memory in the context of Iterable AI?
It’s Iterable AI’s ability to keep a complete, constantly updated profile for every customer by tracking all their past and present interactions across every channel. This gives the AI the context it needs to make sure all future communications are relevant and personalized.
How did the “Momentum Connect” campaign measure customer lifetime value (CLTV)?
We measured CLTV by tracking the total revenue from customers in our targeted segments over the six-month campaign. This included repeat buys, average order value, and subscription renewals, which we then compared to a pre-campaign baseline to find the percentage increase.
What kind of data was integrated into Iterable AI for this campaign?
For this campaign, we fed Iterable AI our first-party data from our CRM, our e-commerce platform (purchase history, abandoned carts), website analytics (browsing, product views), and our customer service ticketing system. This gave us a full picture of every customer’s journey.
Can Iterable AI predict customer churn?
Yes, it uses machine learning to analyze behavior patterns like declining engagement, less feature usage, or no recent purchases. From these patterns, it can predict which customers are a high risk for churning, which lets us try to re-engage them before they leave.
What was the biggest challenge in implementing Iterable AI for this campaign?
Our biggest headache was getting clean and consistent data syncs between all our different internal systems and Iterable. It took a lot of upfront engineering work and required us to build an ongoing data governance process, because the AI is useless without good, accurate data.